Narrative language development of Persian-English bilingual children attending an immersion preschool

Global Perspectives of Bilingualism

Of significant importance are studies from countries and regions around the world to gain a broad perspective on language development. A global perspective is important, particularly because across the regions of the world, multilingualism is viewed differently. In some regions, multiple languages are viewed equitably, multilingualism is encouraged, and students have access to more than one language at school. For example, in Montreal, Canada, French and English have relatively equal status, with bilingual signage around the city and over half of the population being multilingual [5]. While in other regions, one language (e.g., the national language) is promoted with less access to multilingual education. With these differences, variables and contexts should be studied with greater breadth by examining language development across the world.

In the United States, while English is currently the dominant language, linguistic diversity was common in the 1700s and 1800s, as people speaking French, German, Dutch, and Spanish increased. Multilingual education laws were enacted to ensure multilingual education, but by the late 1800s, policymakers began focusing on English as the language that should be taught and used. The Naturalization Act of 1906 made English a requirement for citizenship in the United States, and for the next several decades, more laws were authorized to promote the use and teaching of only English. Since 1906, laws and perceptions have ebbed and flowed, and while other languages do not pose a threat to the English language in the United States and we have learned about the benefits of multilingualism and multilingual education, English-only policies suggest otherwise [6,7]. For example, in 1998, the state of California repealed a multilingual education law, permitting only English instruction in the schools. This was replaced with a law allowing for multilingual education in 2016.

Immersion Programs and Schools

Some countries have a high number of multilingual schools and programs or introduce second languages relatively early. For example, in Sweden, English is introduced in Grade 1, an optional second language in Grade 6, and a third language in Grade 9 [8]. While the second and third languages are optional, it is reported that the majority of students learn a third language (e.g., 70% of 9th grade students in 2017 were also learning French, Spanish, or German) [9]; furthermore, the majority of students (59.3%) with a native language that is not Swedish participate in native language education [10]. In the United States, the number of multilingual programs has increased only recently. In 2000, there were only about 260 dual language programs in the entire United States [11], but by 2015 there were approximately 3,000 programs [12]. However, that number is not sufficient to reach all multilingual children. One study showed that only about 39% of dual language learners in the United States were receiving instruction in their primary language [13]. More recent studies reveal consistent patterns: 3,800 bilingual students requiring bilingual special education in New York City were not placed in bilingual classrooms in 2019 [14].

Multilingualism education has well documented benefits [15]. Multi-language programs (of which there are several different types) promote multilingualism, multiculturalism, and multiliteracy. When students are multilingual and can utilize their family’s heritage language (sometimes referred to as home language, native language, or mother tongue), they are more likely to be closer to family members [16] with greater cohesion and less conflict [17]. When controlling for socioeconomic level, students in multi-language programs have been found to perform academically similarly to, or better than, their monolingual counterparts [18], with no significant disadvantages [19]. However, there is great variability in languages that have been studied and the amount of language input children receive in each language. Thus, the purpose of the current study was to better understand the language development of Persian-English speaking children in the United States attending a Persian immersion preschool and to examine the relationships between language skills and language input.

Persian and Persian-English Bilingualism

Persian is the language most commonly spoken by Iranians in Iran and worldwide but is a less commonly taught language in the United States. It is estimated that the Iranian and Iranian-American population living in the United States includes 476,967 [20] to 1 million people [21], with most people from the United States Census speaking Persian (407,586) [22]. However, despite the growing number of Persian speakers in the United States, little is known regarding the language development of children learning Persian and English in the United States. Some studies have examined very specific language forms of monolingual Persian-speaking children [23–26], but we know of no other studies, aside from our own, that have investigated broader language skills of Persian-English bilingual children in the United States attending a preschool immersion school [27]. The other studies that investigate more specific language forms demonstrate that there are broad similarities in skills and development across languages. For example, Persian-English bilingual children in Canada who demonstrated strong English phonological skills, were stronger at reading in English, similar to their English monolingual peers [23]. Researchers have also found differences in development. For example, a Persian-English bilingual child in Iran developed specific morphemes earlier in Persian than English [28].

Payesteh and Finestack [27] compared Persian and English language performance on standardized assessments of the same group of bilingual Persian-English children included in the present study. Findings indicated that this group of children had stronger performance on assessments of English vocabulary and expressive language than Persian language skills. Additionally, although the assessments used in the study did not include bilingual speakers in their normative samples, the standard scores of the Persian bilingual speakers were all well within the average range. Study results also supported a close relationship between the Persian input children received from their parents and Persian language development, without negative consequence to the development of English. One limitation of the parent study was that it only reported on standardized receptive and expressive assessments of vocabulary and grammar. Thus, the current study extends the scope of the original study by examining the microstructural language (e.g., mean length of utterance, number of different words) derived from English narrative retells of Persian-English bilingual children and comparing their performance to that of monolingual English-speaking children.

Narrative Language in Children

Narrative language is an important aspect of language development in children. In a 15-year longitudinal study [29], narrative ability of children at school entry predicted reading comprehension 10 years later. Researchers have found that error patterns in narratives indicate language impairment in bilingual children [30–35]. Research also indicates that measures of microstructural skills (e.g., number of utterances, syntactic complexity) are strong predictors of language ability in Spanish-English bilingual children [36]. For example, using language samples elicited with wordless picture books, Bedore et al. examined productivity and sentence organization. The productivity measures included number of utterances and number of different words (NDW) produced. The sentence organization measures included mean length of utterance in words (MLU-w) and percentage of grammatical utterances [37,38]. Bedore et al. [36] found that language ability based on the Bilingual English Spanish Assessment [39] was highly correlated with all productivity and organization measures (except number of utterances) in English and Spanish. Moreover, performance on the BESA was best accounted for by English MLU, English grammaticality, and Spanish grammaticality. Because of the close association between these measures of productivity and sentence organization and the BESA, we included each of these microstructural measures in the current study.

Although there is an increasing number of investigations of narrative language development of bilingual children, studies have primarily included Spanish-English bilingual children [36,37,40,41]. Very few studies have examined the narrative language development of other bilingual populations, such as Persian-English speaking children. Other aspects of language development have been examined in monolingual Persian children [42] and Kurdish-Persian speaking [43] children, but there are no studies to our knowledge investigating the narrative skills of Persian-English bilingual children. Thus, this study aimed to explore the English narrative skills of Persian-English bilingual children in the United States and better understand the relationships between language input and development. Specifically, the results of this study will inform speech-language pathologists and other educators regarding the utility of narrative language samples for this population.

Specific Aims

The specific aims (SA) of this study were to:

Participants

This study was approved by the Institutional Review Board at the University of Minnesota and the participating preschools. As part of a larger study, we examined the narrative language skills of two sets of preschoolers. Participants included Persian-English bilingual (BI) children (n=14, M_age=4.1 years, age range: 2;5 (years;months) to 5;7), who attended a Persian immersion preschool. These participants were the same as those included in the Payesteh and Finestack [27] study. The BI children came from homes where, primarily, one parent spoke Persian and one parent spoke English (in this study, each home included a mother and father). About two-thirds of mothers identified as simultaneous Persian-English bilinguals or native-Persian speakers, and just over one-third of fathers identified as simultaneous Persian-English bilinguals or native-Persian speakers. Median household income for all families was above the United States median, 13 of 14 mothers held at least a bachelor’s degree, and 14 of 14 fathers held at least a bachelor’s degree.

The comparison group included monolingual English-speaking (MO) children (n=16, M_age=3.9 years, age range 2;2 to 5;11) who attended an English-only preschool. The MO children came from homes where parents/guardians used English. All but one family had incomes above the United States median, 13 of 16 mothers held at least a bachelor’s degree, and 14 of 16 fathers held at least a bachelor’s degree.

Study Tasks and Procedures

Dependent Measures

Analyses

For the exploration of the BI participants’ narratives skills (Specific Aim 1), microstructural measures were derived using SALT Software; group means and standard deviation were calculated, as well. For the comparisons of the BI and MO groups (Specific Aim 2), two-sample t-tests were conducted. Cohen’s d was calculated to indicate the effect size of each comparison, with large effect size defined as d=0.8, moderate effect size of d=0.5, and small effect size of d=0.2 [54]. Pearson’s correlations were used to examine the relationships across age, English microstructural measures, and language input and use (Specific Aim 3). All data were analyzed in SPSS v25 [55], with p<0.05 to determine statistical significance.

English Narrative Language Skills

To address Specific Aim 1, we examined the narrative microstructural skills for each individual participant and by age group. For each participant, a microstructural narrative profile was developed (Table 1), which included MLU-m, TNW, TNU, NDW, and TTR. A wide range of performance was found across the microstructural measures; however, all participants performed at or above the expected level for their age. Examination of participants’ narrative profiles revealed conformity to the overall group pattern of performance. See Table 1 for group means by age.

Table 1

Microstructure of English Narratives for BI and MO Participants

Group Comparisons

Prior to addressing Specific Aim 2, we determined that there were no statistically significant differences between the BI and MO groups based on socioeconomic status (SES), as measured by income level; and parental education, as measured by the degree obtained by each parent. Next, we compared the narrative performance of the BI participants to that of the MO participants using two-sample t-tests. As shown in Table 2, results revealed no significant differences between the BI and MO groups for four of the five microstructure measures analyzed: TNW, TNU, NDW, and TTR. There was a significant BI and MO group difference based on MLU-m, t=2.4, p<0.05, which was associated with a large effect size (d=0.89). This was characterized by the BI group having a higher average MLU-m (M=6.4, SD=1.74) than the MO group (M=4.9, SD=1.68).

Table 2

Comparison of BI and MO Groups

Factors Related to Narrative Microstructure Measures

To address Specific Aim 3, we examined the associations between our microstructure measures, child age, parental language input, and child language output (Table 3). Each of the microstructure measures was significantly correlated with age. MLU-m, TNW, and NDW were moderately and positively correlated with age (r=.49–.67, p<0.01), while TTR was negatively correlated with age (r=−0.37, p<0.05). Persian parental language input provided by parents had a moderate, positive correlation with MLU-m (r=0.64, p<0.05) and was not significantly correlated with any other microstructure measures. English parental language input provided by parents had a moderate, negative correlation with MLU-m (r=−0.64, p<0.05) and was not significantly correlated with any other microstructure measures. Results revealed that microstructure measures were not correlated with child language production.

Table 3

Correlations between Microstructure Measures and Language Input and Production

English Narrative Skills

Results revealed a wide range of narrative skills, as determined by microstructure measures (Table 1). We note that the BI children produced substantial narratives in English, despite their schooling being entirely in Persian. The MLU-m calculated for each participant’s story retell was within the typical range. For example, Brown’s morphemes dictate that the average 2 ½-year-old should have an MLU-m of approximately between 1.5 and 2.2 [57], and the BI participants in that age range had MLU-m that started at 3.2.

Group Comparisons

When compared with English-speaking monolingual age-matched peers (MO), the BI participants did not perform significantly differently based on TNU, TNW, NDW, and TTR. This indicates that the BI and MO groups were producing narrative story retells that were comparable in terms of their length (i.e., TNU) and lexical diversity (i.e., TNW, NDW, and TTR). Measures of lexical diversity allow us to investigate vocabulary more deeply, with higher ratings of lexical diversity demonstrating a larger vocabulary, as the child is able to use different words throughout their narrative (as opposed to using the same phrases or words). They have also been shown to increase with age and differentiate between disordered and typical language [58,59].

However, there was a significant difference between the two groups when analyzing MLU-m, with the BI group producing an MLU-m (6.4) that was, on average, higher than the MO group (4.9). MLU-m is an average of the length of all of a participant’s utterances and reflects syntactic complexity. Overall, participants in the MO were above or within the expected range for MLU-m, but two MO participants had MLU-m values below the expected range for their age. One explanation for this might be insufficient length of sample. The narratives for these two participants were only 9 and 17 utterances long. Because the recommended length of narratives is at least 50 utterances, we likely did not have enough utterances to derive a reliable MLU-m measure [57]. These language samples of shorter length may also affect the group average, which should be considered when comparing the BI and MO groups. More research with larger sample sizes is required to further interpret our findings.

These results align with studies from other regions of the world, which indicate that children can develop appropriate linguistic skills in the majority language of a country even if their schooling is in a different language [15]. Previous research, in conjunction with the current study, provides positive evidence in encouraging the use of heritage languages for multilingual children as well as supporting meaningful and explicit educational support for those languages.

Factors Related to Narrative Microstructure Measures

There were several notable correlations across our measures. MLU-m, TNW, and NDW each significantly correlated with age, replicating what has been found in previous research [60,61]. This means that as children get older, they produce longer and more complex narratives.

Parental language input in Persian and English significantly correlated with MLU-m, positively and negatively, respectively. This means that children who had a greater amount of Persian spoken to them by their parents (i.e., parental input) tended to have higher MLU-m (English), and children who had greater amounts of English parental input tended to have lower MLU-m (English). As a reminder, nearly all participants (BI and MO) had MLU-m values within the expected range for their age. One explanation for this may be that the parents who used a greater amount of English with their child used less complex language, potentially related to reduced proficiency compared to their Persian language ability. In contrast, children of parents who used more Persian may have received more complex language input, which may have cross-linguistic effects on the child’s English complexity. Another explanation may be based on linguistic differences. Kazemi and Klee [62] found that their group of monolingual Persian children had higher MLU-m than reports of English monolingual children. Persian is a highly inflected language, where a root word could have several morphemes and a single inflected word could create a complete utterance, and a higher MLU-m in English could be an influence of Persian. Further investigation is needed to expand on these results and better determine the role of language input and production in bilingual children in narrative language skills.

Limitations and Future Directions

The following limitations should be considered when interpreting the results of the study. First, the small sample size makes it challenging to observe statistical significance and to determine if any one participant is an outlier or simply represents one end of a range. The specific population studied resulted in a small pool from which to recruit. Second, the story retell task was difficult for some younger participants, thus reducing the number of participants available for analyses. Including a greater number of older preschool-aged participants who were more successful with the task would have been beneficial or utilizing a task better suited to the younger children where performance could be compared across tasks. Third, many participants did not produce the recommended minimum number of utterances in their language samples which could have affected the microstructure measures. Fourth, parent questionnaires were not always completed in their entirety or at all, thus limiting our ability to use participant data when analyzing their task performance with parental language input and child language production. A shorter parent questionnaire with more focused questions or one where parents may complete the questionnaire in different formats (e.g., paper, electronic) may allow for greater completion.

Concluding Remarks

This study adds to a small body of research on Persian-English bilingual children and children who attend language immersion preschools. Findings demonstrate that Persian-English bilingual children who attend a Persian immersion preschool in the United States are capable of learning English to a level comparable to same-age English monolingual peers and can display their language skills in a story retell task. Studies such as this are important for the progress of education, and speech-language pathology more specifically. First, they provide support for multilingualism, helping us extrapolate what has been found with other languages and language pairs to those languages that are less commonly studied. Moreover, researchers can build upon research with typically developing populations to populations with language disorders. This also allows speech-language pathologists to utilize the research to inform practice with culturally and linguistically diverse populations, particularly in differentiating language disorders from language differences. Finally, studies such as this can be used to inform policy regarding cultural and linguistic diversity, meaning we can promote linguistic diversity and equity by endorsing multilingualism and multi-language immersion programs, with the knowledge that children can be successful at communicating in more than one language.